[SOLVED] Two Node Cluster on VE6.1

cl4x

Member
Dec 17, 2019
6
0
6
42
Hi,
i have only 2 node cluster and i see this strange behavior: when one node dead, all VM shutdown. I don't understand why the VE compromise my health infrastructure on surviving node.

Ok, no problem, with pvecm e 1 i recovery the VM (but with a fail), so I looked for a way to properly configure a 2-node cluster, so I found this guide: https://pve.proxmox.com/wiki/Two-Node_High_Availability_Cluster#Quorum_Disk_Configuration . I have a SAN attached with iSCSI and this isnt a problem for me.

But i think this guide is old, there are no way to execute the command mkqdisk. This command is unreacheable in any repository (cman does not exist and the alternate fence-agents and dlm-controld don't have mkqdisk). How i format the quorum disk? There are another way to set permanently the command pvecm e 1 and remove the quorum limitation?

Thanks in advance.
 
hi,

i have only 2 node cluster and i see this strange behavior: when one node dead, all VM shutdown. I don't understand why the VE compromise my health infrastructure on surviving node.

when you have a cluster with only 2 nodes, you will lose quorum when one of them goes offline. when you lose quorum, cluster cannot operate until quorum is restored.

There are another way to set permanently the command pvecm e 1 and remove the quorum limitation?

you can add a quorum device[0] to your cluster setup, so it would be a 2+1 cluster. quorum device gives an extra vote, and therefore you can take one of your nodes offline without the cluster losing quorum.


[0]: https://pve.proxmox.com/pve-docs/chapter-pvecm.html#_corosync_external_vote_support
 
Hellou,
can I ask for help with similar conditions? I have two nodes too, but without shared storage. So for me is good, to set expected votes only 1 forever. Can it be done without external vote? On each node are separate vms, with own data. Sometimes one node is down for many reasons and users on proxmox cannot operate with other vms, only when administrator set pvecm expected 1.
Thank you for anwser.
 
Hellou,
can I ask for help with similar conditions? I have two nodes too, but without shared storage. So for me is good, to set expected votes only 1 forever. Can it be done without external vote? On each node are separate vms, with own data. Sometimes one node is down for many reasons and users on proxmox cannot operate with other vms, only when administrator set pvecm expected 1.
Thank you for anwser.
Code:
# vim /etc/corosync/corosync.conf
quorum {
    provider: corosync_votequorum
    two_node: 1
    wait_for_all: 0
}
 
Hi, in a similar situation, however is there any set Hardware Requirements for a cheap Quorum QDevice for now? I only want the third Node just to have quorum. no VMs no nothing just make the decision if Node 1 or Node 2 Fail. Does this hardware have to be VM capable? I was looking at just getting one of these to go with my x3950 x6.

How To: Proxmox VE Lenovo Nano Cluster